Colorectal adenomatous polyps are precursors of

Transcription

1 GASTROENTEROLOGY 2010;139: The Presence of Large Serrated Polyps Increases Risk for Colorectal Cancer SAKIKO HIRAOKA,* JUN KATO,* SHIGEATSU FUJIKI, EISUKE KAJI,*,, TAMIYA MORIKAWA, TAKATOSHI MURAKAMI, # TORU NAWA,** MOTOAKI KURIYAMA,* TOSHIO URAOKA,* NOBUYA OHARA, and KAZUHIDE YAMAMOTO* *Department of Gastroenterology and Hepatology, Okayama University Graduate School of Medicine, Dentistry, and Pharmaceutical Sciences, Okayama; Tsuyama Central Hospital, Okayama; Japanese Red Cross Society Himeji Hospital, Hyogo; Mitoyo General Hospital, Kagawa; Fukuyama Medecal Center, Hiroshima; # Saiseikai Imabari Hospital, Ehime; **Fukuyama City Hospital, Hiroshima; and Department of Pathology and Oncology, Okayama University Graduate School of Medicine, Dentistry, and Pharmaceutical Sciences, Okayama, Japan This article has an accompanying continuing medical education activity on page e12. Learning Objective: Upon completion of this exam regarding serrated polyps, successful learners will be able to evaluate the risk of colorectal cancer in patients having serrated polyps, and to determine appropriate ways to follow-up the patients with such polyps. See editorial on page BACKGROUND & AIMS: There is evidence that serrated polyps (serrated adenomas and hyperplastic polyps) have different malignant potential than traditional adenomas. We used a colonoscopy database to determine the association between the presence of serrated colorectal polyps and colorectal neoplasia. METHODS: We performed a multicenter observational study of 10,199 subjects who underwent first-time colonoscopies. Data collected on study subjects included age and sex and the location, size, and histology of polyps or tumors found at colonoscopy. Serrated polyps were defined as those diagnosed by the pathologists in the participating hospitals as a serrated lesion (a lesion given the term of classical hyperplastic polyp, traditional serrated adenoma, sessile serrated adenoma, or mixed serrated polyp ). Large serrated polyps (LSPs) were defined as those 10 mm. RESULTS: There were 1573 patients (15.4%) with advanced neoplasia, 708 patients (6.9%) with colorectal cancer (CRC), and 140 patients (1.4%) with LSPs in our cohort. Multivariate analysis associated the presence of LSPs with advanced neoplasia (odds ratio [OR], 4.01; 95% confidence interval [CI], ) and CRC (OR, 3.34; 95% CI, ). The presence of LSPs was the greatest risk factor for CRC, particularly for proximal CRC (OR, 4.79; 95% CI, ). Proximal and protruded LSPs were the highest risk factors for proximal CRC (OR, 5.36; 95% CI, and OR, 9.00; 95% CI, , respectively). CONCLUSIONS: The presence of LSPs is a risk factor for CRC, particularly CRC of the proximal colon. Keywords: Advanced Neoplasia; Colon Cancer; Colon Cancer Risk. Colorectal adenomatous polyps are precursors of most colorectal cancer (CRC). In contrast, all hyperplastic polyps (HPs) traditionally have been considered to have little malignant potential. Histologically, HPs are serrated polyps that are characterized by a sawtooth architecture. In 2003, Torlakovic et al classified serrated polyps according to histologic features as HPs, traditional serrated adenomas (TSAs), and sessile serrated adenomas (SSAs). 1 Recent studies have reported that some serrated polyps are precursors of CRC (serrated pathway). 1 4 In particular, SSAs have unique characteristics with regard to their clinical and genetic background, such as a proximal dominant location, high frequency of BRAF mutations, 3,4 and a CpG island methylator phenotype (CIMP) (CpG island hypermethylation). 5 Therefore, SSAs have been claimed to represent precursor lesions for CRC showing microsatellite instability-high (MSI-H). 2 However, it is yet to be elucidated whether patients with serrated polyps are at risk for developing colorectal neoplasia. A few reports have described the relationship between HPs and development of CRC. 6 8 In examinations of the correlation of the occurrence of HPs with the development of CRC, hyperplastic polyposis syndrome (HPS) has been given attention as has familial adenomatous polyposis (FAP) in studies of the adenoma-carcinoma sequence. HPS is a rare syndrome defined by the presence of numerous HPs and is associated with an approxi- Abbreviations used in this paper: CI, confidence interval; CIMP, CpG island methylator phenotype; CRC, colorectal cancer; FAP, familial adenomatous polyposis; HP, hyperplastic polyp; HPS, hyperplastic poplyposis syndrome; LSP, large serrated polyp; MP, mixed serrated polyp; MSI-H, microsatellite instability-high; OR, odds ratio; SSA, sessile serrated adenoma; TSA, traditional serrated adenoma by the AGA Institute /$36.00 doi: /j.gastro

2 1504 HIRAOKA ET AL GASTROENTEROLOGY Vol. 139, No. 5 mately 50% lifetime risk of developing CRC. 2,9,10 As to sporadic cases of serrated polyps, Glazer et al reported in a study of a series of 17 cases with serrated adenomas that these patients were more likely to develop adenomatous polyps than control patients. 11 Recently, Li et al showed that large serrated polyps ( 10 mm) (LSPs) were strongly and independently associated with synchronous advanced colorectal neoplasia among 4714 asymptomatic subjects. 12 The report appears to be the first to show a correlation between serrated polyps and development of adenomatous polyps in a large cohort. However, their subjects were asymptomatic and included only 467 patients with advanced neoplasia and 33 patients with invasive cancer. Therefore, the relationship between LSPs and development of CRCs was not shown in that study. Accordingly, in this study, we examined a large multicenter colonoscopy database to evaluate the relationship between serrated polyps and the occurrence of colorectal advanced neoplasia, particularly CRC. Moreover, the differences in this relationship according to characteristics of polyps or cancer (ie, location, number, and morphology) were also investigated. Materials and Methods Study Design We have maintained a multicenter colonoscopy database since This database includes clinical data on all patients who underwent colonoscopy in Okayama University Hospital and 14 affiliated hospitals. Data collected include age; gender; indications for colonoscopy; and location, size, and histology of polyps or cancer found at colonoscopy. Data input was performed by each colonoscopist or, in a few institutes, by nurses or assistants. From June 2005 to May 2008, results of a total of 33,962 colonoscopies were registered into this database. Among these, 11,570 first-time colonoscopy cases were initially considered eligible for this study. Exclusion criteria included cases less than 20 years of age, with prior resection of any part of the colon, with FAP or hereditary nonpolyposis colorectal cancer, with inflammatory bowel disease, and whose data lacked clinical information or histologic information on polyps. After excluding these subjects, 10,199 cases were analyzed. This database analysis was approved by the Institutional Review Board of Okayama University Hospital and each affiliated hospital. All potential investigator conflicts of interest were disclosed to study participants. Informed consent was obtained from each patient. Colonoscopic Findings During the colonoscopy, the location and size of all polypoid lesions and cancers were determined. Polyp or cancer locations were classified into 2 groups: proximal, defined as the cecum, ascending colon, and transverse colon (including the splenic flexure); and distal, defined as the descending colon, sigmoid colon, and rectum. Size of the polyps resected with polypectomy or surgical resection was measured just after resection. The polyps that underwent biopsy alone were measured endoscopically using standerd clinical practice, for example, the visual estimation and the open biopsy forceps method. 13,14 Macroscopic appearance of the polyp was classified according to the Paris endoscopic classification, 15 and data were input as follows: protruded or pedunculated (Ip), intermediate or broad-based forms (semipedunculated [Isp], sessile [Is], and superficially elevated [IIa]) (Supplementary Figure 1). In the analysis of morphology of LSP, Ip or Isp were regarded as protruded, and Is or IIa were considered to be the flat type. The precise estimation of polyp size and macroscopic appearance was of particular importance in this study. To minimize intra- or interobserver variability, therefore, the photos of all polyps observed by colonoscopy were reviewed afterwards at the conferences, which were held at each hospital once a week with attendance of all endoscopists including specialists and trainees. Consensus regarding the polyp size and macroscopic appearance was usually achieved during the process of evaluations of reports written by each endoscopist. However, the opinions of the most experienced specialists at each hospital generally had the greatest impact on the final decision. Pathologic Findings Histologic studies were performed on all polyps and tumors taken by polypectomy or biopsy procedures by board certified pathologists at each participating hospital. CRCs included all invasive cancer and cancer in Tis category (carcinoma in situ and intramucosal) of the TNM clinical classification. 16 Traditional tubular adenomas were classified as tubular, tubulovillous, or villous adenomas. They were also classified as low grade or high grade according to the grade of dysplasia. Adenomas 10 mm, with a villous component, or with high-grade dysplasia or cancer were defined as advanced neoplasia. Serrated polyps were classified as HPs, SSAs, mixed serrated polyps (MPs), or TSAs using the system described in detail by Torlakovic et al. 1 Briefly, HPs were lesions that deviated minimally from normal with only mild serration and prominent goblet cells alternating with absorptive-type columnar cells (goblet cell type) or were lesions with increased serration and columnar cells that were distended with small, apical vesicles (microvesicular type). SSAs were characterized by exaggerated serration, architectural changes including crypt dilatation and horizontally extended crypt bases, increased mucin production (both intracellular and intraluminal), increased proliferation, and mild nuclear changes falling short of dysplasia. MPs comprised nondysplastic HPs or SSAs and had a dysplastic component that resembled

3 November 2010 LARGE SERRATED POLYPS AND COLORECTAL CANCER 1505 Statistical Analysis We used 2 test or Fisher s exact test for univariate analysis and logistic regression model for multivariate analysis to identify factors contributing to advanced neoplasia or CRC. The multivariate analyses were performed including all variables used in the corresponding univariate analyses. Regarding interaction by side of the colon, we performed a statistical test of difference for proximal and distal colon using the 2 test. This analysis was performed excluding patients with advanced neoplasia (or CRC) in both sides of the colon or without any advanced neoplasia (or CRC). Results were shown with calculations of odds ratios (ORs) and 95% confidence intervals (CIs). These analyses were performed using the JMP program (version 8; SAS Institute, Cary, NC). All P values were 2 sided and considered significant when less than.05. Results Clinical Characteristics of Patients Of the 11,570 persons enrolled in this study, 10,199 cases were eligible for analysis. Excluded were those who had insufficient pathologic data (n 1181), Table 1. Characteristics of Study Population No. of patients (%) Figure 1. Macroscopic and histological appearance of LSPs observed with chromoendoscopy. Representative endoscopic images and histologic features of (A) hyperplastic polyp, (B) SSA, (C) TSA, and (D) MP. either TSAs or conventional adenomas. TSAs had epithelium that was unequivocally dysplastic or adenomatous throughout as well as a serrated architecture. For convenience of data input, histology of all types of serrated polyps was input as hyperplastic or serrated adenoma. If pathologists assigned the diagnosis of HP, the polyp was input as hyperplastic. If the diagnosis was TSA or MP, the polyp was input as serrated adenoma. SSA was input as either hyperplastic or serrated adenoma according to details of the pathologic review. LSPs were polyps that fulfilled the following 2 categories in the database: 10 mm and histology indicating hyperplastic or serrated adenoma. Figure 1 shows images of the typical macroscopic appearance and histologic features of LSPs. Variable Subjects at enrollment (n 11,570) Eligible subjects (n 10,199) Sex Male 6096 (52.7) 5257 (51.5) Female 5473 (47.3) 4942 (48.5) Unknown 1 Age Mean standard deviation, (y) y 462 (4.0) 342 (3.4) y 996 (8.6) 956 (9.4) y 1475 (12.7) 1351 (13.2) y 2552 (22,1) 2304 (22.6) y 2823 (24.4) 2459 (24.1) y 2473 (21.4) 2114 (20.7) 80 y 778 (6.7) 673 (6.6) Unknown 11 Indication of colonoscopy Screening 1019 (8.9) 874 (8.6) Positive fecal occult blood test 4771 (41.4) 4229 (41.5) Abdominal symptoms 2214 (19.1) 2106 (20.6) Rectal bleeding 1453 (12.3) 1351 (13.2) Anemia 248 (2.2) 241 (2.4) Others 1271 (11.0) 896 (8.8) Unknown 594 (5.2) 502 (4.9) Neoplasia No ademona 7112 (69.7) Ademona 9 mm a 1514 (14.8) Advanced neoplasia 1573 (15.4) Ademona 10 mm 450 (4.4) Tubulovillous or villous 216 (2.1) ademona High-grade dysplasia b 199 (2.0) Cancer 708 (6.9) LSPs 140 (1.4) a Except adenomas with high-grade dysplasia. b Except adenomas with adenoma 10 mm.

4 1506 HIRAOKA ET AL GASTROENTEROLOGY Vol. 139, No. 5 Table 2. Analysis of Factors Predictive of Advanced Neoplasia Subjects without advanced neoplasia (n 8626) Subjects with advanced neoplasia (n 1573) Univariate analysis, Multivariate analysis, Age y y ( ) b 1.96 ( ) b Gender Female Male ( ) b 1.61 ( ) b Number of small ademonas a ( ) b 5.12 ( ) b LSPs No Yes ( ) b 4.01 ( ) b NOTE. Multiple analysis includes age, gender, number of small ademonas, and the presence of large serrated polyps. a Excluding histologically advanced ademonas. b P were less than 20 years of age (n 94), had prior resection of any part of the colon (n 21), had FAP (n 2), had inflammatory bowel disease (n 64), and had insufficient clinical data (n 9). Characteristics of the study population are shown in Table 1. The subjects included 4942 (48.5%) women and 5257 (51.5%) men, with a mean age of 58.9 years. Indications for colonoscopy were screening (874, 8.6%), positive fecal occult blood test (4229, 41.5%), abdominal symptoms (2106, 20.6%), rectal bleeding (1351, 13.2%), anemia (241, 2.4%), and other reasons. There were 1573 (15.4%) patients with advanced neoplasia in our cohort. Among the subjects were 450 (4.4%) with large tubular adenoma (diameter 10 mm), 216 (2.1%) with villous or tubulovillous adenoma, 199 (2.0%) with high-grade dysplasia, and 708 (6.9%) with CRC. In addition, 140 (1.4%) patients had LSPs. Predictors of Colorectal Advanced Neoplasia and Cancer Analyses of factors predictive of advanced neoplasia and CRC are shown in Tables 2 and 3. By univariate analysis, all of the potential risk factors examined (advanced neoplasia: age, male sex, number of small adenomas, and presence of LSPs; CRC: age, number of adenomas, presence of large adenomas [ 10 mm], and presence of LSPs) except sex in case of CRC were statistically significant. In addition, all significant factors shown by univariate analysis were also found to be significant by multivariate analysis. For advanced neoplasia, Table 3. Analysis of Factors Predictive of Colorectal Cancer Subjects without cancer (n 9491) Subjects with cancer (n 708) Univariate analysis, Multivariate analysis, Age y y ( ) a 2.63 ( ) a Gender Female Male ( ) 1.05 ( ) Number of ademonas ( ) a 1.65 ( ) c Adenomas 10 mm No Yes ( ) a 1.56 ( ) b LSPs No Yes ( ) a 3.34 ( ) a NOTE. Multiple analysis includes age, gender, number of adenomas, the presence of adenomas 10 mm and the presence of large serrated polyps. a P b P.001. c P.01.

5 November 2010 LARGE SERRATED POLYPS AND COLORECTAL CANCER 1507 age 65 years (OR, 1.96; 95% CI, ), male sex (OR, 1.61; 95% CI, ), number of small adenomas 4 (OR, 5.12; 95% CI, ), and presence of LSPs (OR, 4.01; 95% CI, ) were all significant, whereas, for CRC, age 65 years (OR, 2.63; 95% CI, ), number of adenomas 4 (OR, 1.65; 95% CI, ), presence of adenoma 10 mm (OR, 1.56; 95% CI, ), and presence of LSPs (OR, 3.34; 95% CI, ) were significant factors. Among these risk factors, the highest OR for the risk of CRC was the presence of LSPs. Because this analysis largely depended on the pathologic interpretation, we performed multivariate analysis including hospital as a covariate in the logistic model. In these analyses, the results did not differ significantly from the results in the initial analysis shown in Tables 2 and 3. The ORs of presence of LSPs were 3.97 (95% CI, ) for advanced neoplasia and 3.35 (95% CI, ) for cancer, respectively. We also performed multivariate analysis in 5 hospitals with more than 1000 subjects separately, showing that the trend was not much skewed by each hospital (Supplementary Table 1). Table 4. Multivariate Analysis of Factors Predictive of Advanced Neoplasia According to Location Variables Distal advanced neoplasia, Proximal advanced neoplasia, Age ( ) b 2.05 ( ) b Male gender 1.72 ( ) b 1.34 ( ) c Number of small 3.74 ( ) b 4.85 ( ) b ademonas 4 a LSPs 2.96 ( ) b 3.63 ( ) b NOTE. Multiple analysis includes age, gender, number of small adenomas, and the presence of large serrated polyps. a Excluding advanced ademonas. b P c P.01. Table 5. Multivariate Analysis of Factors Predictive of Colorectal Cancer According to Location Variables Distal cancer, Proximal cancers, Age ( ) a 3.73 ( ) a Male gender 1.32 ( ) b 0.62 ( ) c Number of small 1.91 ( ) b 1.24 ( ) ademonas 4 Adenomas 10 mm 1.38 ( ) b 1.95 ( ) b LSPs 2.23 ( ) b 4.79 ( ) a NOTE. Multiple analysis includes age, gender, number of adenomas, the presence of adenomas 10 mm and the presence of large serrated polyps. a P b P.01. c P.001. Association Between LSPs and Synchronous Advanced Colorectal Neoplasia or CRC According to Location Because clinical and biologic differences have been shown recently to exist between colorectal neoplasia in the proximal and distal colon, 17,18 we next analyzed the predictors of colorectal advanced neoplasia and cancer according to location (Tables 4 and 5). These analyses revealed that the risk presented by the presence of LSPs was equal for advanced neoplasia and cancer either in the proximal or distal colon (advanced neoplasia: distal [OR, 2.96; 95% CI, ], proximal [OR, 3.63; 95% CI, ]; CRC: distal [OR, 2.23; 95% CI, ), proximal [OR, 4.79; 95% CI, ]). We also made analysis without either carcinoma in situ or intramucosal cancer (ie, including only CRCs with submucosal or deeper invasion) and showed the results in Supplementary Table 2. Similar results were also obtained in this analysis, although the statistical significance of the correlation of LSP with distal cancer was lost. Thus, the presence of LSPs showed the highest OR as the risk factor for proximal CRC. In this context, we calculated the OR of distal neoplasia (advanced neoplasia or CRC) to proximal neoplasia on each variable (Supplementary Table 3). The presence of LSPs was more influential to proximal CRC than distal CRC (OR of distal CRC to proximal CRC, 0.48; 95% CI: ). Analysis According to Characteristics of LSPs Finally, we examined the associations between the characteristics of LSPs and risk of distal or proximal colorectal neoplasia. LSPs were classified as proximal, distal, flat type (including sessile), protruded, and solitary or multiple. Relative risks for each category are shown in Table 6. The presence of distal LSPs and flat-type LSPs are significant risk factors for both proximal and distal colorectal advanced neoplasia and cancer (distal LSP: OR, 2.57; 95% CI, for distal advanced neoplasia; OR, 2.93; 95% CI, for proximal advanced neoplasia; OR, 2.44; 95% CI, for distal cancer; OR, 4.12; 95% CI, for proximal cancer; flat-type LSP: OR, 3.97; 95% CI, for distal advanced neoplasia; OR, 2.99; 95% CI, for proximal advanced neoplasia; OR, 2.74; 95% CI, for distal cancer; OR, 2.66; 95% CI, for proximal cancer). In contrast, proximal LSPs and protruded LSPs are associated mainly with the risk of proximal neoplasia (proximal LSPs: proximal advanced neoplasia OR, 4.25; 95% CI, ; proximal cancer OR, 5.36; 95% CI, ; protruded LSPs: proximal advanced neoplasia OR, 5.24; 95% CI, ; proximal cancer OR, 9.00; 95% CI, ). The only risk associated with multiple LSPs was that of distal advanced neoplasia (OR, 3.42; 95% CI, ). Thus, most LSPs have a potent association with the occurrence of any type of advanced neoplasia or cancer. Moreover, specific correlations were observed be-

6 1508 HIRAOKA ET AL GASTROENTEROLOGY Vol. 139, No. 5 Table 6. Characteristics of Large Serrated Polyps and Risk of Colorectal Neoplasia Distal advanced neoplasia, adjusted a Proximal advanced neoplasia, adjusted a Distal cancer, adjusted b Proximal cancer, adjusted b Proximal LSP 3.41 ( ) c 4.25 ( ) d 2.10 ( ) 5.36 ( ) d Distal LSP 2.57 ( ) c 2.93 ( ) c 2.44 ( ) e 4.12 ( ) c Flat-type LSP 3.97 ( ) d 2.99 ( ) d 2.74 ( ) c 2.66 ( ) e Protruded LSP 1.62 ( ) 5.24 ( ) d 1.75 ( ) 9.00 ( ) d Number of LSP ( ) e 2.70 ( ) 3.15 ( ) 4.65 ( ) a Relative risk adjusted for age, gender, and number of small adenomas. b Relative risk adjusted for age, gender, number of adenomas, and the presence of adenoma ( 10 mm). c P.01. d P.001. e P.05. tween proximal or protruded LSPs and proximal colorectal neoplasia. Discussion In this large population that underwent colonoscopy, we found that the presence of LSPs was an independent risk factor for the presence of advanced neoplasia, particularly CRC. In addition, among the potential risk factors for colorectal neoplasia, the presence of LSPs correlated most significantly with the occurrence of proximal CRC. Moreover, subanalysis according to characteristics of LSPs revealed that proximal LSPs and protruded LSPs were particularly correlated with proximal CRC. To the best of our knowledge, only the report of Li et al has shown a correlation between LSPs and colorectal neoplasia using a large cohort that underwent screening colonoscopy. 12 Although their study had the advantage of minimizing bias by using an asymptomatic cohort, it also has the inevitable drawback that the proportion of patients with neoplasia, particularly CRC, was relatively small. Therefore, although they could show a correlation between LSPs and advanced neoplasia, they could not show a correlation between LSPs and CRC. In contrast, our study cohort was composed of subjects who underwent colonoscopy for a variety of indications. The substantial advantage of using this cohort is that the proportion of patients with colorectal neoplasia was relatively large. In fact, the proportion of subjects with CRC was as high as 6.9%, and, therefore, we could definitely show a positive correlation between the presence of LSPs and the occurrence of CRC. Thus, our results showed novel aspects of the malignant potential of LSPs and risk of cancer in patients with LSPs. In addition, the large percentage of those with CRC in our cohort enabled us to undertake several subanalyses regarding the effects of location, number, and morphology of LSPs on the correlation with CRC. Recent consensus has indicated that serrated polyps have several different histologic types: HP, SSA, TSA, and MP. 1 HPs are small, usually 5 mm in diameter, symmetrical, and uniform. Epithelial serration is confined to the surface and upper part of the crypts, and proliferation occurs at the crypt base. SSAs are slightly elevated lesions with irregular borders and may be covered with mucus. TSAs have a predilection for the distal colorectum but are distributed throughout the proximal colon. TSAs usually have a pedunculated or broad-based polypoid growth pattern with a pineal-form endoscopic appearance. 19 MPs comprise nondysplastic HPs or SSAs and a dysplastic component that resembles either TSAs or conventional adenomas. Thus, there is more than 1 category of serrated polyps, and those in our database included all types according to the diagnosis of pathologists at each hospital. The precise association between each type of serrated polyp and CRC occurrence could not be determined in this study and is expected to be shown in further studies. In molecular terms, there are several possible explanations for the correlation between CRC and LSPs. First, LSPs may have a molecular background similar to certain types of CRC and therefore have the potential to progress to CRC. Accumulating histologic and molecular evidence suggests that certain serrated polyps, including TSAs, SSAs, and possibly some classic HPs, are precursors of CRC, particularly MSI-H CRC, 20 which accounts for approximately 10% of all sporadic CRC. 21 Notably, molecular alterations in SSA are common with those of MSI-H cancer, such as CpG island hypermethylation, 5 decreased expression of hmlh1 or hmsh2, 1 and BRAF mutations. 2 4 In addition, the microvesicular subtype of HPs has a high rate of BRAF mutations, whereas K-ras mutations are more common in the goblet cell type HPs, 2,3 raising the possibility that microvesicular HPs may be precursors for SSAs but leaving open the possibility that the goblet cell type of HPs has malignant potential as well. 2 Thus, the results of this study may reflect a common genetic property between serrated polyps and concomitant CRC and reinforce the theory that LSPs are precursors of CRC. However, it has not yet been confirmed that LSPs actually progress into CRC. In fact, CRCs that develop in HPS patients are largely microsatellite stable rather than MSI-H, suggesting that serrated polyps do not always progress into MSI-H cancer. 2,9,22 On

7 November 2010 LARGE SERRATED POLYPS AND COLORECTAL CANCER 1509 this issue, further clinical and molecular studies are required. Second, LSPs could be a marker of the cancer field. It has recently been postulated that genetic and epigenetic changes in normal colonic mucosa may be correlated with the occurrence and development of colorectal neoplasia (field cancerization). 23 In particular, we and other groups have indicated that epigenetic changes such as DNA methylation in normal colonic mucosa may be an early event in the occurrence of some forms of colorectal neoplasia. 24,25 Regarding the epigenetic changes in normal colonic mucosa of patients with serrated polyps, there are some forms of HPS in which the earliest manifestation is hypermethylation of multiple gene promoters. 26 Therefore, serrated polyps and concomitant CRC may arise from the same (epigenetic) field of cancerization. The present study included a subanalysis of differences in the correlation of CRC according to LSPs with various characteristics. This aspect has not been the subject of a previous report. We found differences according to location of both LSPs and concomitant CRC and the macroscopic appearance of LSPs. In particular, proximal LSPs were correlated with the occurrence of proximal neoplasia, particularly proximal CRC. This result suggests the correlation between serrated polyps and proximal CRC and may be a manifestation of common molecular backgrounds between them, such as CIMP, BRAF mutation, and MSI-H because these genetic and epigenetic alterations are specific to proximal CRC. 21,27 To confirm this hypothesis, however, the results should be validated in other populations. Another notable finding is that protruded LSPs were strongly correlated with proximal cancer but not with distal cancer. Protruded LSPs mainly include TSAs and some of SSAs and MPs, although the precise proportions of these pathologies could not be determined. In contrast, most HPs are flat-type LSPs. It has been shown that TSA and SSA are likely to have the BRAF mutation and exhibit CIMP, whereas HPs sometimes have K-ras mutations (never BRAF mutations) and occasionally exhibit CIMP. 20 These molecular manifestations exactly reflect the molecular differences between proximal and distal cancer. Thus, our subanalysis is consistent with what is known about the molecular biology of colorectal neoplasia and may show important clinical aspects regarding the serrated pathway of colorectal carcinogenesis. It was previously reported that the prevalence of LSPs was 2.3% of subjects studied, 12 whereas we found LSPs in 1.4% of subjects. There may be difference in the frequency according to characteristics of cohort, such as demographics and ethnicity. Uncertain generalizability to other populations regarding prevalence of LSPs and relationship to neoplasia is a limitation of this study. Because LSPs are likely to be precursors of MSI-H cancer, however, the relative proportion of CRCs (6.9%) to LSPs (1.4%) in our study may reflect clinical aspects in general that MSI-H cancer accounts for 10% 15% of sporadic CRCs. There are other limitations to this study. First, some polyps may have had insufficient histologic evaluation of biopsy specimens because MPs have the component of traditional adenoma. Therefore, some of these polyps might not have been classified as serrated polyps. Second, there were differences in the detection rate of serrated polyps, especially SSA and HPs, among participating hospitals. This may be due to the differences in the skill of the endoscopists because experts are likely to differentiate HPs from adenomas without a biopsy using a variety of endoscopic modalities, such as magnification endoscopy and narrow band imaging. 28,29 Third, because of database analysis, we could not classify serrated polyps into HP, SSA, MP, and TSA. However, the precise distinction between HP and SSA is difficult, and it has been reported that the agreement rate among pathologists is low. 30 Fourth, the results of relationship between subtypes of LSPs and advanced neoplasia/crc were still exploratory and need to be validated in further analysis. Finally, the genetic information on CRC such as CIMP, MSI status, and BRAF and K-ras mutations was unavailable. Therefore, it is uncertain whether LSPs and concomitant CRC harbor common molecular backgrounds. Our results suggest that LSPs are associated with CRC and advanced neoplasia, particularly proximal advanced neoplasia and CRC. Moreover, proximal and protruded LSPs have a strong association with proximal advanced neoplasia and CRC. These results certainly reflect genetic backgrounds of colorectal tumorigenesis involving serrated polyps that were correlated with CIMP and MSI-H. Patients with LSPs require meticulous follow-up with colonoscopy, and prospective observation of these patients may reveal novel findings regarding development of proximal CRC or CRC with MSI. Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at and at doi: /j.gastro References 1. Torlakovic E, Skovlund E, Snover DC, et al. Morphologic reappraisal of serrated colorectal polyps. Am J Surg Pathol 2003;27: Mäkinen MJ. Colorectal serrated adenocarcinoma. Histopathology 2007;50: Spring KJ, Zhao ZZ, Karamatic R, et al. High prevalence of sessile serrated adenomas with BRAF mutations: a prospective study of patients undergoing colonoscopy. Gastroenterology 2006;131: Minoo P, Moyer MP, Jass JR. Role of BRAF-V600E in the serrated pathway of colorectal tumourigenesis. J Pathol 2007;212:

8 1510 HIRAOKA ET AL GASTROENTEROLOGY Vol. 139, No Wynter CV, Walsh MD, Higuchi T, et al. Methylation patterns define two types of hyperplastic polyp associated with colorectal cancer. Gut 2004;53: Otori K, Oda Y, Sugiyama K, et al. High frequency of K-ras mutations in human colorectal hyperplastic polyps. Gut 1997;40: Jass JR, Young J, Leggett BA. Hyperplastic polyps and DNA microsatellite unstable cancers of the colorectum. Histopathology. 2000;37: Jass JR, Whitehall VL, Young J, et al. Emerging concepts in colorectal neoplasia. Gastroenterology 2002;123: Rashid A, Houlihan PS, Booker S, et al. Phenotypic and molecular characteristics of hyperplastic polyposis. Gastroenterology 2000; 119: Boparai KS, Mathus-Vliegen EM, Koornstra JJ, et al. Increased colorectal cancer risk during follow-up in patients with hyperplastic polyposis syndrome: a multicentre cohort study. Gut 2010; 59: Glazer E, Golla V, Forman R, et al. Serrated adenoma is a risk factor for subsequent adenomatous polyps. Dig Dis Sci 2008; 53: Li D, Jin C, McCulloch C et al. Association of large serrated polyps with synchronous advanced colorectal neoplasia. Am J Gastroenterol 2009;104: Park SH, Choi EK, Lee SS, et al. Polyp measurement reliability, accuracy, and discrepancy: optical colonoscopy versus CT colonography with pig colonic specimens. Radiology 2007;244: Gopalswamy N, Shenoy VN, Choudhry U, et al. Is in vivo measurement of size of polyps during colonoscopy accurate? Gastrointest Endosc. 1997;46: Participants in the Paris Workshop. The Paris endoscopic classification of superficial neoplastic lesions: esophagus, stomach, and colon: November 30 to December 1, Gastrointest Endosc 2003;58(6 Suppl):S Colon and Rectum. In: Sobin LH, Wittekind C, eds. International Union Against Cancer, TNM Classification of Malignant Tumours. 6th ed. New York: Wiley-Liss, 2002: Komuro K, Tada M, Tamoto E, et al. Right- and left-sided colorectal cancers display distinct expression profiles and the anatomical stratification allows a high accuracy prediction of lymph node metastasis. J Surg Res 2005;124: Nawa T, Kato J, Kawamoto H, et al. Differences between rightand left-sided colon cancer in patient characteristics, cancer morphology and histology. J Gastroenterol Hepatol 2008;23: Morita T, Tamura S, Miyazaki J, Higashidani Y, Onishi S, et al. Evaluation of endoscopic and histopathological features of serrated adenoma of the colon. Endoscopy 2001;33: O Brien MJ, Yang S, Mack C, et al. Comparison of microsatellite instability, CpG island methylation phenotype, BRAF and KRAS status in serrated polyps and traditional adenomas indicates separate pathways to distinct colorectal carcinoma end points. Am J Surg Pathol 2006;30: Samowitz WS, Albertsen H, Herrick J, et al. Evaluation of a large, population-based sample supports a CpG island methylator phenotype in colon cancer. Gastroenterology 2005;129: Rubio CA, Stemme S, Jaramillo E, et al. Hyperplastic polyposis coli syndrome and colorectal carcinoma. Endoscopy. 2006;38: Ushijima T. Epigenetic field for cancerization. J Biochem Mol Biol 2007;40: Hiraoka S, Kato J, Horii J, et al. Methylation status of normal background mucosa is correlated with occurrence and development of neoplasia in the distal colon. Hum Pathol 2010;41: Kawakami K, Ruszkiewicz A, Bennett G, et al. DNA hypermethylation in the normal colonic mucosa of patients with colorectal cancer. Br J Cancer 2006;94: Minoo P, Baker K, Goswami R, et al. Extensive DNA methylation in normal colorectal mucosa in hyperplastic polyposis. Gut 2006; 55: Koinuma K, Shitoh K, Miyakura Y, et al. Mutations of BRAF are associated with extensive hmlh1 promoter methylation in sporadic colorectal carcinomas. Int J Cancer 2004;108: Fu KI, Sano Y, Kato S, Fujii T, et al. Chromoendoscopy using indigo carmine dye spraying with magnifying observation is the most reliable method for differential diagnosis between nonneoplastic and neoplastic colorectal lesions: a prospective study. Endoscopy 2004;36: Machida H, Sano Y, Hamamoto Y, et al. Narrow-band imaging in the diagnosis of colorectal mucosal lesions: a pilot study. Endoscopy 2004;36: Farris AB, Misdraji J, Srivastava A, et al. Sessile serrated adenoma: challenging discrimination from other serrated colonic polyps. Am J Surg Pathol 2008;32: Received February 4, Accepted July 7, Reprint requests Address requests for reprints to: Jun Kato, MD, Department of Gastroenterology and Hepatology, Okayama University Graduate School of Medicine, Dentistry, and Pharmaceutical Sciences, Shikata-cho, Kita-ku, Okayama , Japan. katojun@ cc.okayama-u.ac.jp; fax: (81) Acknowledgments The authors thank Mayumi Tokumitsu for her invaluable help in data input and Dr Junji Shiode (Okayama Saiseikai General Hospital), Dr Masahiro Takatani (Japanese Red Cross Society Himeji Hospital), Dr Jiro Miyaike (Saiseikai Imabari Hospital), Dr Tatsuya Toyokawa (Mitoyo General Hospital), Dr Masashi Tatsukawa (Onomichi Municipal Hospital), Dr Seiyuu Suzuki (Sumitomo Besshi Hospital), Dr Fumiya Yoshinaga (Japanese Red Cross Society Mihara Hospital), Dr Yoshio Sakane (Mizushima Central Hospital), Dr Takashi Nakamura (Takahashi Central Hospital), Dr Jun Tomoda (Fukuyama Medecal Center), Dr Sou Tsuboi (Shigei Medical Reseach Hospital), Dr Hideyuki Suzuki (Okayama University Hospital), and Dr Reiji Higashi (Okayama University Hospital) for their help with the database. Conflicts of interest The authors disclose no conflicts.

9 November 2010 LARGE SERRATED POLYPS AND COLORECTAL CANCER 1510.e1 Supplemental Figure 1. Macroscopic appearance of LSP corresponding each category of the Paris polyp classification. Endoscopic pictures and schematic images of: (A) superficially elevated, IIa, (B) sessile, Is, (C) semipedunculated, Isp, and (D) pedunculated, Ip.

10 1510.e2 HIRAOKA ET AL GASTROENTEROLOGY Vol. 139, No. 5 Supplemental Table 1. Multivariate Analysis of Factors Predictive of CRC Stratified by Hospital Hospital Subject, n LSP, n Cancer, n Age 65, Male gender, Number of ademonas 4, Ademonas 10 mm, LSPs, A ( ) a 1.35 ( ) 3.73 ( ) c 1.33 ( ) 4.89 ( ) b B ( ) a 1.03 ( ) 1.37 ( ) 1.28 ( ) 3.88 ( ) d C ( ) b 0.95 ( ) 1.87 ( ) 2.02 ( ) d 3.23 ( ) d D ( ) a 1.22 ( ) 1.50 ( ) 1.00 ( ) 2.15 ( ) E ( ) a 0.93 ( ) 2.16 ( ) 2.21 ( ) 2.22 ( ) a P b P.01. c P.001. d P.05. Supplemental Table 2. Multivariate Analysis of Factors Predictive of CRC Including Only CRCs With Submucosal or Deeper Invasion Variables All CRC, Distal CRC, Proximal CRC, Age ( ) a 2.25 ( ) a 3.91 ( ) a Male gender 0.92 ( ) 1.21 ( ) 0.50 ( ) a Number of ademonas ( ) 1.28 ( ) 0.59 ( ) Adenomas 10 mm 1.57 ( ) b 1.47 ( ) d 1.80 ( ) d LSPs 2.86 ( ) c 1.73 ( ) 4.93 ( ) a NOTE. Multiple analysis includes age, gender, number of adenomas, the presence of adenomas 10 mm, and the presence of LSPs. a P b P.01. c P.001. d P.05.

11 November 2010 LARGE SERRATED POLYPS AND COLORECTAL CANCER 1510.e3 Supplemental Table 3. OR of Distal Neoplasia to Proximal Neoplasia on Each Variable Variables Advanced neoplasia, Cancer, Age ( ) 0.57 ( ) c Male gender 1.33 ( ) b 2.10 ( ) d Number of small 0.74 ( ) ND adenomas 4 a Number of adenomas 4 ND 1.50 ( ) Adenomas 10 ND 0.81 ( ) LSPs 0.73 ( ) 0.48 ( ) b ND, not determined. a Excluding advanced adenomas. b P.05. c P.01. d P.0001.